Bourbaki

Exercise 58, § 1

Commutative Algebra, IX, p. 68

a) Let ε(T)\varepsilon(T) be the formal series exp⁡(−∑n=0∞Tpn/pn)\exp(- \sum_{n=0}^\infty T^{p^n}/p^n) of Q[[T]]\mathbf{Q}[[T]]. We have

ε(T)=∏(n,p)=1n integer ≥1(1−Tn)μ(n)/n \varepsilon(T) = \prod_{\substack{(n,p)=1 \\ n \text{ integer } \geq 1}} (1 - T^n)^{\mu(n)/n}
where μ\mu is the Möbius function. The coefficients of ε(T)\varepsilon(T) belong to Z(p)\mathbf{Z}_{(p)}. b) Let X=(Xn)n∈NX = (X_n)_{n \in \mathbf{N}} be a family of indeterminates. In the algebra Q[(Xn)n∈N]\mathbf{Q}[(X_n)_{n \in \mathbf{N}}], we have the equality
∏n=0∞ε(XnTpn)=exp⁡(−∑n=0∞Φn(X)Tpn/pn). \prod_{n=0}^\infty \varepsilon(X_n T^{p^n}) = \exp(- \sum_{n=0}^\infty \Phi_n(X) T^{p^n}/p^n).
c) Let AA be a Z(p)\mathbf{Z}_{(p)}-algebra. Suppose that multiplication by p.1Ap.1_A is injective in AA. Let a=(an)n∈N∈ANa = (a_n)_{n \in \mathbf{N}} \in A^\mathbf{N}. For the series exp⁡(∑n=0∞anTpn/pn)\exp(\sum_{n=0}^\infty a_n T^{p^n}/p^n) of A[1p][[T]]A\left[\frac{1}{p}\right][[T]] to have its coefficients in AA, it is necessary and sufficient that aa belong to ΦA(W(A))\Phi_A(W(A)). d) Let AA be a Z(p)\mathbf{Z}_{(p)}-algebra. The mapping E∘ωAE \circ \omega_A of W(A)W(A) into Λ(A)\Lambda(A) (cf. p. 54, Exercise 40 and p. 57, Exercise 43) associates with a=(an)n∈Na = (a_n)_{n \in \mathbf{N}} the element ∏n=0∞ε(an(−T)pn)\prod_{n=0}^\infty \varepsilon(a_n(-T)^{p^n}) of Λ(A)\Lambda(A). e) Let AA be a Z(p)\mathbf{Z}_{(p)}-algebra. Suppose that multiplication by p.1Ap.1_A is injective in AA. Let σ\sigma be an endomorphism of AA satisfying σa≡apmod  pA\sigma a \equiv a^p \mod pA. Let sσ:A→W(A)s_\sigma : A \to W(A) be the homomorphism associated with σ\sigma (p. 44, Exercise 14). Then, for every a∈Aa \in A, the formal series
Eσ(a,T)=exp⁡(∑i=0∞σi(a) Tpi/pi) E_\sigma(a, T) = \exp(\sum_{i=0}^\infty \sigma^i(a)\ T^{p^i}/p^i)
has its coefficients in AA, and we have
E∘ωA∘sσ(a−1)=exp⁡(∑i=0∞σi(a) (−T)pi/pi)=Eσ(a,−T). E \circ \omega_A \circ s_\sigma(a^{-1}) = \exp(\sum_{i=0}^\infty \sigma^i(a)\ (-T)^{p^i}/p^i) = E_\sigma(a, -T).
f) Let A\mathscr{A} be the ring Z(p)[(Xn)n∈N]\mathbf{Z}_{(p)}[(X_n)_{n \in \mathbf{N}}] and let X=(Xn)n∈N∈W(A)X = (X_n)_{n \in \mathbf{N}} \in W(\mathscr{A}). The series EFA(X,T)E_{F_{\mathscr{A}}}(X, T) has its coefficients in W(A)W(\mathscr{A}). For every Z(p)\mathbf{Z}_{(p)}-algebra AA and every element a=(an)n∈Na = (a_n)_{n \in \mathbf{N}} of W(A)W(A), we denote by EF(a,T)E_F(a, T) the series obtained by applying W(φ)W(\varphi) to the coefficients of EFA(X,T)E_{F_{\mathscr{A}}}(X, T), where φ:A→A\varphi : \mathscr{A} \to A is the mapping which associates ana_n with XnX_n. Then EF(a,T)E_F(a, T) has its coefficients in W(A)W(A) and if sAs_A denotes the homomorphism W(A)→W(W(A))W(A) \to W(W(A)) defined in Exercise 15, p. 44, we have
E∘ωW(A)∘sA(a−1)=EF(a,−T)for alla∈W(A). E \circ \omega_{W(A)} \circ s_A(a^{-1}) = E_F(a, -T) \quad \text{for all} \quad a \in W(A).

Solution

No solution has been written yet.

Machine translation of the English, by gpt-5.4. Not checked by a person.